DeepMind leader frames long view

Demis Hassabis used a new interview published April 7 to describe Google DeepMind as a lab still organized around very long-range questions about intelligence, power and human survival. (singjupost.com) The interview was hosted by Cleo Abram and posted as “Huge Conversations,” with Abram telling viewers she would ask what Hassabis plans to do with the power he now holds across Google’s artificial intelligence work. Hassabis is the chief executive of Google DeepMind, the company formed after Google combined DeepMind and Google Brain in April 2023. (singjupost.com) (deepmind.google) Google DeepMind says its mission is to “build AI responsibly to benefit humanity,” and its public careers pages describe research engineers as people who combine software engineering, mathematics and research skills to advance that mission. The company’s hiring language stresses long-term problems, large-scale systems and work tied to artificial general intelligence. (deepmind.google 1) (deepmind.google 2) (deepmind.google 3) That framing matches Hassabis’s public message over the past year. In a Time interview published April 27, 2025, he said he had worked on artificial general intelligence for his entire career and argued it could help tackle disease, energy and climate if built “properly and responsibly.” (time.com) Hassabis has tied that long view to scientific status as well as commercial scale. He won the 2024 Nobel Prize in Chemistry with John Jumper for AlphaFold, the protein-structure system that Time described as having cut work that once took years down to hours for many researchers. (nobelprize.org) (time.com) AlphaFold matters here because it is a narrow system: it predicts protein shapes from amino-acid sequences, but it does not reason across the wider world. Hassabis told Time that his larger goal is still broader artificial intelligence that can move beyond single tasks and help make discoveries itself. (time.com) He has also kept the safety argument in view. Google DeepMind wrote on April 2, 2025, that artificial general intelligence could arrive “within the coming years” and said the lab was prioritizing technical safety, risk assessment and outside collaboration. (deepmind.google) That stance has drawn scrutiny as Google’s role in defense and government work has grown. Time reported in 2025 that Google now sells services, including DeepMind systems, to militaries, including the United States, after an earlier DeepMind pledge against military and weapons use had fallen away. (time.com) Hassabis’s interview this week did not settle that tension. It showed the same through line he has repeated across Nobel, Time and Google DeepMind appearances: the lab wants people who can build hard systems now while keeping their eyes on questions measured in decades. (singjupost.com) (nobelprize.org) (deepmind.google)